Nature examines the science behind AI extinction claims

Nature's news team examined the recent wave of AI-extinction warnings and found the case, in the words of one interviewed researcher, closer to faith than science.

The September 22 explainer by Elizabeth Gibney opens on Jacob Coxon, who resigned from Anthropic on September 8 saying he feared the company's AI systems "could spiral out of control and destroy humanity." His post drew over 100 million views in 24 hours. Evan Hubinger, who leads Anthropic's alignment science efforts, put the risk of human extinction at ">10% within the next decade." Anthropic CEO Dario Amodei called for a slowdown, not a halt, in AI development, and Sam Altman and Elon Musk backed him.

The piece walks through the mechanisms proponents invoke, then hands the microphone to skeptics. RAND's Michael Vermeer told Nature that extinction predictions involve "so many untestable claims that you just end up with a conversation that is really more like faith than something scientific." A 2025 RAND analysis of nuclear, biotech and atmospheric-modification scenarios ruled out nuclear extinction but could not rule out the other two, though they would require AI systems with "considerable ability to physically interact with the world."

Heidy Khlaaf of the AI Now Institute told the magazine that "AI's low reliability and accuracy rates in critical environments" concerned her more than extinction, and she called existential-risk messaging "fear-mongering." Nearly 1,400 AI sector employees have signed an open letter demanding a slowdown, and David Sacks of the U.S. President's Council of Advisors on Science and Technology warned that companies face "massive product-liability exposure" from cyberattacks their systems could enable.

The Nature explainer moved through the researcher circles we track within a day of publication.